The Human Connectome Project (HCP) was launched by the National Institutes of Health in 2009 to build a circuit diagram of the anatomical and functional connectivity of the human brain. The hope was to produce a body of data to facilitate research into brain disorders such as dyslexia, autism, Alzheimer's disease, and schizophrenia. Based on the principles of connectomics, scientists have already successfully constructed a network map of a mouse's retina and primary visual cortex.
The ultimate goal is to understand how the human brain processes sensory information and represents knowledge, and to use this knowledge to implement such processes in the design of intelligent devices. The SyNAPSE project, funded by DARPA and jointly conducted by a consortium of universities and IBM computer labs, strives to put the computing power of a human brain on a silicon chip no larger than two liters in volume. The ambition is to build a microprocessor with 10 billion artificial "neurons" and 100 trillion "synapses," effectively replicating one hemisphere of a human brain.
The project, if successful, will result from a culmination of 30 years of research on simulated neural networks. Called "neuromorphic processing," the computing activities performed by the neural processors could solve problems too complex for conventional computers. According to team leader Dharmendra Modha, such neuromorphic processing will not replace conventional computing, but enhance it; for example, through advanced pattern recognition, such as the ability to identify a single face in a crowd.
An intelligent system, whether biological or electronic, must be capable of performing four basic activities. First, it must be able to perceive its environment. Second, it must have memory in which to store its perceptions. Third, it must be capable of applying its perceptions to formulate behavior and evaluate the effectiveness of its actions. Fourth, it must be able to autonomously communicate with other intelligent systems.